Market Integration and Energy Trade Efficiency: An Application of Malmqvist Index to Analyse Multi-Product Trade

Transcription

1 Chapter 6 Market Integration and Energy Trade Efficiency: An Application of Malmqvist Index to Analyse Multi-Product Trade Yu Sheng Australian National University Yanrui Wu University of Western Australia Xunpeng Shi National University of Singapore Dandan Zhang Peking University September 2014 This chapter should be cited as Sheng, Y., Y. Wu, X. Shi and D. Zhang (2014), Market Integration and Energy Trade Efficiency: An Application of Malmqvist Index to Analyse Multi-Product Trade, in Han, P. and F. Kimura (eds.), Energy Market Integration in East Asia: Energy Trade, Cross Border Electricity, and Price Mechanism, ERIA Research Project Report FY2013, No.29.Jakarta: ERIA, pp

2 CHAPTER 6 Market Integration and Energy Trade Efficiency: An Application of Malmqvist Index to Analyse Multi-Product Trade YU SHENG 1 Australian National University YANRUI WU, University of Western Australia XUNPENG SHI, National University of Singapore DANDAN ZHANG Peking University This paper uses the data envelope analysis method to investigate the Malmquist indexbased gravity relationship between bilateral energy trade flows and their determinants throughout the world. Using a balance panel data of 40 countries between 1995 and 2008, this paper shows that market integration will increase energy trade by improving trade efficiency between trade partners, though allowing for a flexible substitution between different energy products tends to weaken these effects. This result highlights crossproduct substitution and its implications for the aggregate energy trade pattern, providing insights on the importance of prioritising product-specific trade facilitating policies. Keywords: energy trade efficiency, energy market integration, Malmquist index JEL Classification: Q27, Q47, O47 1 Corresponding author: yusheng@gmail.com, Research Fellow, the Australian National University 135

3 Introduction Rapid economic growth in East Asia has substantially affected global energy consumption and its pattern over the past three decades. Between 1980 and 2012, the average gross domestic product (GDP) growth rate of countries in this region is more than 5 percent a year, which is more than double the GDP growth of Organisation for Economic Co-operation and Development (OECD) countries for the same period. The sustained economic growth, mainly due to the rapid expansion of manufacturing industries, led to two consequences. On one hand, it generated a huge increase in energy demand in the region and throughout the world. On the other hand, it created a significant disparity in energy supply and demand across regions. Since the late 1980s, energy consumption growth in this region has accounted for more than two-thirds of the world total, and the cumulative energy demand by this region is still increasing and likely to reach between 7 billion and 8 billion tonnes of oil equivalent (btoe) by 2030 (IEA, 2012). Scholars and policy makers have reached the consensus that facilitating crosscountry energy trade through forming a more integrated regional or global energy market can help stabilize market prices for energy products and secure energy supply (Shi and Kimura, 2010; Wu, et al., 2014). This is because moving toward a more integrated energy market will increase the allocation efficiency of limited energy resources and resolve many economic and political issues related to the imbalance between energy supply and demand. However, limited progress has been made in practice, particularly from developing countries perspectives. An important reason is that the aggregate benefits that all participants could obtain from involving themselves into regional and global market integration for energy products are hard to justify. In addition, there are also concerns about the fairness of benefit allocation across countries. To quantify trade creation effects an important benefit from forming market integration trade economists have long been using the gravity model to examine the relationship between bilateral trade flow and its determinants (Anderson, 1979; Anderson and Wincoop, 2003; Costinot and Rodriguez- Clare, 2013). In literature, an essential argument is that market integration can increase trade efficiency and thus improve the welfare of all trade partners by 136

4 providing additional trade creation. For example, Rose (2004) used a gravity model with a large panel data that covered over 50 years and 175 countries, and this showed that joining the Generalised System of Preferences (GSP) raised the bilateral trade by 136 percent, while Subramanian and Wei (2007) showed that membership to the General Agreement on Tariffs and Trade/World Trade Organization (GATT/WTO) significantly increased imports (around 44% of world trade) for industrial countries though unevenly across countries.applying this method to analyse the impact of market integration on energy trade creation, many studies (Sheng and Shi, 2013) have also found a substantial positive trade creation effects through joining a more integrated energy market. Although previous studies contribute to improve general knowledge, the accuracy of their predictions on the trade creation effects of market integration has always been criticized. In particular, the predicted trade creation or trade efficiency obtained from using the data at different aggregation levels are always inconsistent to each other (Subramanian and Wei, 2007). A possible explanation for this phenomenon, among others, is that the standard gravity model usually uses the aggregate trade value (i.e., summed up from commodities) as the dependent variable for the regression analysis. This treatment simplifies the exercise, but neglects the potential role of substitution/complementarity between various trade components in affecting the aggregate bilateral trade flow. This paper uses the Malmquist index approach a method initially designed for estimating the multi-output and multi-input production function to investigate the gravity relationship between bilateral energy trade flows and their determinants. In contrast to previous studies, the approach used in this paper allows for a flexible substitution between different energy products in bilateral trade and thus provide a better measure of trade creation and trade efficiency due to energy market integration (EMI). Using a balance panel data for 40 countries between 1995 and 2008, this paper shows that regional integration will generally increase trade creation and trade efficiency though its effects on different products are different. Compared to the conventional gravity model with perfect cross-product substitution, results in this paper suggest that the substitution between 137

5 different energy products is likely to weaken the aggregate trade creation effects (or the trade efficiency gain) due to market integration. Moreover, the implicit shadow price of specific energy products relative to others (derived from the simulation) can change over time, implying that cross-product substitution and market integration process is interacted. A policy implication is that policy makers aiming to promote the bilateral energy trade flow need to prioritise the trade of the most valuable energy products. The remainder of the paper is organized as detailed below. Section 2 discusses the changing pattern of global energy trade and its components over the past two decades. A brief summary of the related literature follows. Section 3 provides the methodology and estimation strategy. The Malmquist index approach is employed to examine the gravity relationship between the bilateral energy trade flows and their determinants, and to provide the measure of trade efficiency when allowing for a flexible substitution between different energy products in trade. Section 4 describes the variables to be used and the related data sources, and provides descriptive statistics. Section 5 discusses the empirical results and Section 6 presents the conclusions. Global Energy Trade and Cross-Product Substitution The energy trade has grown rapidly throughout the world over the past two decades, though its growth pattern is unevenly distributed across regions (Figure 6.1). Between 1995 and 2008, the total value of energy trade throughout the world has increased from US$249.5 million (at constant 2005 prices) to US$ million with an annual growth rate of 16.8 percent. The growth in energy trade associated with countries in the East Asia Summit (EAS) region is the most important driver. The total value of energy trade among the EAS countries and between the EAS countries and the rest of the world has increased from US$28 million and US$123 million, respectively, in 1980 to US$132 million and US$657 million in When added together, these account for around 70 percent of total world energy trade. Along with the strong growth in total energy flow, trade pattern has also become more diversified. The number of pairs trade has increased from 991 to 1,271 between 1995 and

6 Figure 6.1: Global energy trade and its components, by region, A) Total trade flow and the number of pairs trade, (in US$ billion at 2005 prices) B) Cross-region distribution of energy trade, (in US$ 000 at 2005 prices) Source: Global Trade Analysis Project (GTAP) Energy Dataset. However, the strong growth in total energy trade does not evenly apply to all energy products (Figure 6.2). Over the period , oil trade has been dominating the total energy trade. The average proportion of oil in total energy trade is around 80 percent, followed by natural gas (11%) and coal (9%). In terms of growth, the growth of trade in natural gas has taken the lead with an average annual growth rate of 19 percent, followed by oil trade (16.8%) and coal trade (13.5%). The uneven proportion (in total trade) and 139

7 growth of trade in different energy products reflect their relative importance in the bilateral energy trade. Figure 6.2: Components of global energy trade, by products, (in US$ billion at 2005 prices) Source: Global Trade Analysis Project (GTAP) Energy Dataset. The relative importance of different products also varies across different regions (Figure 6.3). For example, more than one-fourth of energy trade between countries within the EAS region is trade in coal and its share in total regional energy trade has increased from 26 percent in 1995 to 38 percent in In contrast, trade in coal only accounted for 7 percent of total energy trade between the EAS countries and the rest of world in 1995 and its share has further declined to less than 4 percent in The disparity in the relative importance of different products across regions is not only determined by the trading partners characteristics in resource endowments, consumption preference, and production capacities but is also affected by the ease of different trade components substitutability in consumption and its dynamic changes. Failing to consider this latter point may generate biased estimates on the aggregate trade flow. 140

8 Figure 6.3: Cross-region comparison of energy trade components, (in US$ million at 2005 prices) A) Energy trade between the EAS countries, by products, B) Energy trade between the EAS countries and the rest of the world, by products, Source: Global Trade Analysis Project (GTAP) Energy Dataset. Although there have been a large number of studies exploring the gravity relationship between bilateral energy trade and its determinants, only quite a few attempts have been made to combine the gravity model (for explaining 141

9 the relationship between bilateral trade flow and its determinants) with the stochastic frontier analysis or the data envelope analysis(originally designed to measure efficiency in production or cost functions (Kuosmanen, et al. 2004) to quantify trade efficiency and its potential trade creation effects due to market integration. Trade efficiency is defined as the distance between actual trade flows and the maximum trade possible. Following earlier studies in this field, several works (Drysdale and Garnaut, 1982; Kalirajan, 1999; Kalirajan and Findlay, 2005; Kang and Fratianni, 2006) applied the stochastic frontier analysis to the standard gravity model and investigated trade efficiency across 10 groups of countries throughout the world between 1975 and 2000 by using the bilateral trade data sets from Ross (2004). They showed that developed countries generally had higher trade efficiency than developing countries, and global and regional market integration contributed to raise cross-country trade efficiency. Among the Asia-Pacific region, the ASEAN has the highest trade efficiency while South Asian countries have the lowest efficiencies. Kalirajan (1999) and Miankhel, et al. (2009) used the same method to examine the trade efficiency between Australia and its 65 trading partners during They found that China and Japan, as well as ASEAN countries, are the key major trading partners that could provide substantial potential for Australia s trade in mineral products (including energy products). Kalirajan and Singh (2008), following Drysdale, et al. (2000), examined the trade efficiency between China and its 56 trading partners and found that China s efficiency was higher for trade with other Asia-Pacific region economies (especially, Chile, Hong Kong, Indonesia, Malaysia, Singapore, and Thailand) than with the European Union (EU) and the United States (US). Roperto (2013) and Roperto and Edgardo (2014) examined the trade efficiency between the Philippines and its trade partners and found that global and regional integration tend to increase trade efficiency among ASEAN countries. The existing literature, though providing some useful information, suffers in general with two shortcomings. First, most of these studies focused on total trade with little implication for bilateral energy trade and the related market integration policies. Second, like conventional gravity studies, most of these 142

10 researches use aggregate trade value as dependent variable to measure trade efficiency, which neglected the effects of cross-product non-substitution. In this paper, the Malmquist index is used to measure efficiency of multiproduct energy trade when flexible substitution between trade components is considered. The Malmqvist Index Approach and Trade Efficiency Measure When investigating the gravity relationship between bilateral trade flow and its determinants, one can start by using a standard empirical specification, initially derived by Anderson (1979) and Anderson and Van Wincoop (2003), such that where is the exporting country, is the importing country, and is the industry (or commodity/commodity group). (1) The terms and are income levels, which vary only at the and levels. captures the partial equilibrium effects of bilateral trade barrier or trade policies. is the residual that is used to capture the randomly distributed unobserved white noises. Equation (1) can be estimated by using different methodologies for specific purposes, including the identification of bilateral trade determination, the assessment of negative effects of regional integration, and so on. In the literature for measuring trade efficiency, the stochastic frontier analysis or the data envelopment analysis are usually employed for the regression. Specifically, one can retrieve the best performing trade flow given trading partners income level, trade barriers, and other controlled factors, and compare it with other trade flows to quantify their relative differences as a measure of trade efficiency. Normally, Equation (1) is specified to take the 143

11 constant elasticity of substitution (CES) or the trans-log forms, and is assumed to contain an inefficient component ( <0) and a white noise ( ), such that. These methods work well for analysing trade flow ( ) at the commodity level, but it could not provide useful information on how trade flow may evolve and whether they are efficient at the aggregate level. This is because the substitution/complementary relationship between different components can usually change their aggregation and thus affect the measure of trade pattern at the aggregate level and its corresponding trade efficiency. In particular, when there are no perfect substitution between trade components, the model may tend to overestimate potential trade flow and trade efficiency. To deal with the multi-outcome case, productivity economists designed the distance function method to retrieve the real substitutive/complementary relationship between different outputs (i.e., in production function), namely the Malmquist index. The method, initially used for estimating the production function, can now be used to investigate the gravity relationship between multi-product bilateral energy trade and its determinants. Since it assumes a relatively more flexible conversion function between different energy trade components, changes of trade in each energy product between any pair of trading partners can be identified through the calculation of the relative ratio of the distance of each data point relative to a commonly shared potential frontier. With the standard assumption of imperfect substitution between multiproduct energy trades ( ) and between trade determinants (, the Malmquist index between period and is given by: (2) This index is estimated as the geometric mean of two distance functions: one used as a reference the potential trade frontier at period t and the other used as a reference at period t+1 (Fare et al., 1994). Since the reference point can be defined as the potential maximum trade flow that could be achieved once the related trade determinants are constant, the Malmquist index can be treated as a measure of trade efficiency relative to the reference and its change over 144

12 time could provide information on how the trade efficiency changed over time. Moreover, Fare et al. (1994) also showed that the Malmquist index could be decomposed into an efficiency change component and a technical change component, and that these results could be applied to the different periodbased Malmquist indexes. (3) The efficiency change component of the Malmquist indexes measures the change in how far the observed trade is from the maximum potential trade between period t and t+1, and the technical change component reflects the shift of natural created trade (due to demand and preferences) between the two periods. To define the trade determinants-based Malmquist index, it is necessary to characterise the trade determination mechanism (namely, the gravity model) and estimate its efficiency in trade generation. Using Equations (2) and (3), the trade creation mechanism describes the possibilities for the transformation of trade determinants (x t ->R+) such as GDP, bilateral distances, and trade policies into energy trade flows (y t ->R+). Yet, the method looks like a black box and could not directly provide the relative importance of the different energy products as components in the total bilateral energy trade. To deal with this problem, this paper followed Coelli and Rao (2001) by using the simulation method and deriving the implicit share (or marginal contribution of various trade components and trade determinants) in the Malmquist index following the neoclassical assumption. All efficient possibilities of bilateral energy trade in the time period t is characterised by the set (or the frontier of the set) of The technology satisfies the usual set of axioms: closeness, non-emptiness, scarcity, and no free lunch. The frontier of the set for a given output vector is defined as the input vector that cannot be decreased by a uniform factor (4) 145

13 without leaving the set. Such a frontier can be estimated by using a minimisation process s.t. k=1,,m j=1,,n (5) where represents the r different TUs that defined the trade frontier, k are m trade flows, and j are n trade determinants. The efficiency score obtained ( ) will take values between 0 and 1, with indicating that the bilateral trade is located at the frontier. Equation (5) is known as the data envelop form of the approach. An equivalent dual approach can be derived from its primal form (Kuosmanen, et al., 2004). The envelope approach is preferred to the distance function way for estimating trade efficiency since it requires fewer constraints. Also, the current form has the advantage of a more intuitive specification, offering a better economic interpretation of the problem. Using the above method, the impact of EMI policies on trade creation of multi-products can be estimated at the same time. In particular, the marginal contribution of each product to various determinants to trade can be isolated from the others through the dual method. This provides some useful knowledge to inform the relevant policies, since the marginal contribution of various trade determinants can be converted into corresponding cost-benefit ratios. Data Collection and Variables Definition Data used in this study come from four major sources including (i) the global trade analysis project (GTAP) energy product database, (ii) the UN Comtrade 146

14 Database and data used by Subramanian and Wei (Subramanian and Wei, 2007), (iii) the World Development Indicator Database, (iv) and the energy statistics from the BP Statistical Review of World Energy. Initially, the database cover the bilateral trade in three types of energy products, including coal, petrol, and gas across 172 countries (including 26 EAS countries) over the period Yet, the real number of trade flows is much smaller than the initial dataset and many trade flows are zeros. This is because energy trade across countries heavily depends on exporting countries initial natural endowments. Since the gravity model is more reliable in providing long-term projection, this paper uses the five-year average to smooth the year-to-year fluctuation in energy trade. Finally, the estimation of Malmquist index requires the balanced panel data, which impose the additional constraints. With all three constraints considered, the sample size is cut down to 1,164 pairs of bilateral trade, covering 40 countries over four time periods 1995, 2000, 2005, and The sample is representative since they are added up to account for 44 percent of total energy trade of the whole world in 2008, which include 60 percent of coal trade, 43 percent of oil trade, and 45 percent of natural gas trade. The dependent variable the bilateral trade in coal, petrol, and gas between each pair of countries is defined as real import value of each commodity. To make it comparable across countries and over time, nominal import values are deflated by using the corresponding commodity price at 2005 prices (provided by the GTAP datasets). It is to be noted that the import value rather than the total trade value was deliberately used to represent the bilateral trade since energy trade is usually a one-way trade. With such a treatment, the bilateral energy trade can be better captured by the characteristics of importers and exporters. Independent variables first include the GDP per capita of both importers and exporters in US dollars at constant 2000 price and the geographical distances between the corresponding trade partners. Data for the period are coming directly from Subramanian and Wei (2007) while data for the period after 2000 are coming from the World Development Indicator Database. Some adjustments have been made to make them consistent over time. In addition to the standard variables used in gravity models, the natural endowment of energy products in exporting countries are also used as control 147

15 variables. This is important since it is impossible for countries holding no natural reserve in energy products to export. Data on natural endowment of natural reserves of each type of energy products in exporting and importing countries are obtained from various issues of the BP Statistical Review of World Energy. Table 6.1 provides the summary statistics of the dependent variables (the bilateral trade in three energy products) and the major independent variables (i.e., GDP per capita, distance, and natural reserve in individual energy products). Table 6.1: Logarithm of major variables in the regression Variable Names No. of Obs. Mean Std. Dev. Min. Max. ln_agg._energy_trade ln_coal_trade ln_oil_trade ln_gas_trade ln_gdp_capita_importer ln_gdp_capita_exporter ln_distance ln_land_area dummy for common language dummy for FTA share of manufacturing industry ratio of energy to non-energy trade coal_ reserve_importer oil_reserve_importer gas_reserve_importer coal_ reserve_exporter oil_reserve_exporter gas_reserve_exporter Note: FTA = Free Trade Agreement, GDP = gross domestic product, No. of Obs. = Number of observations., Std. Dev. = standard deviation, max. = maximum, min. = minimum. Source: Global Trade Analysis Project (GTAP) Energy Dataset. 148

16 Empirical Results: Multi-Product Energy Trade Determinants and Its Efficiency Bilateral Trade Determination and Substitution between Trade Components Applying the Malmquist index method to the data of bilateral energy trade, the gravity relationship is estimated between bilateral energy trade flows and their determinants, including the trading partners economic growth, trade barriers (i.e., distance) and other controlled variables such as country-specific industrial trade and structure, Free Trade Agreement (FTA) participation, and initial endowment in natural resources. For robustness check, results obtained from two models are compared. The first model only uses the trading partners GDP per capita and the geographical distance as the determinants of bilateral energy trade while the second model also incorporates other controlled variables. The results are shown in Table 6.2. When allowing for more flexible substation/complementarities between different energy products, the marginal contribution of various trade determinants to bilateral trade flows are measured and reported in Table 6.2. These results are further compared with those obtained from the model, which uses the aggregate energy trade flow as the dependent variable. 149

17 Table 6.2: Marginal Contribution of Trade Determinants to the Aggregate Energy Trade Single- Product Energy Trade Model I Multi-Product Energy Trade Single- Product Energy Trade Model II Multi-Product Energy Trade ln_gdp_per_capita_importer 0.035*** 0.009** 0.040*** 0.011** (0.005) (0.004) (0.005) (0.004) ln_gdp_per_capita_exporter 0.025*** 0.019*** 0.037*** 0.027*** (0.006) (0.005) (0.006) (0.005) ln_distance *** ** (0.008) (0.006) (0.009) (0.007) Ratio of energy to non-energy trade * 0.964** - - (0.512) (0.412) Share of secondary industry in GDP *** 0.724*** - - (0.015) (0.103) Dummy_for_FTA ** (0.022) (0.017) coal_reserve_cty *** *** - - (0.000) (0.000) oil_reserve_cty *** 0.001*** - - (0.000) (0.000) gas_reserve_cty *** 0.007*** - - (0.002) (0.001) coal_reserve_cty *** 0.000*** - - (0.000) (0.000) oil_reserve_cty *** 0.000* - - (0.000) (0.000) gas_reserve_cty (0.004) (0.003) Constant 0.559*** 0.443*** 0.664*** 0.496*** (0.105) (0.084) (0.114) (0.089) Note: FTA = Free Trade Agreement, GDP = gross domestic product. *** p<0.01, ** p<0.05, * p<0.1 Source: Global Trade Analysis Project (GTAP) Energy Dataset. Consistent with the prediction of conventional gravity models, trading partners economic growth positively contributed to bilateral energy trade while geographical distance negatively contributed to bilateral energy trade (Table 6.2). However, the magnitude of these coefficients of trade determinants is much smaller than that obtained from the traditional models (which assume that different energy products are perfectly substituted). This implies that using the aggregate energy trade flow as dependent variable may 150

18 tend to overestimate the potential trade driven by conventional gravity drivers and thus cause the overestimation of trade efficiency, which is defined as the gap of real trade flow relative to potential trade flow. As an example, Table 6.3 compares the average growth in efficiency of bilateral energy trade between using the sum of energy trade (or the singleproduct trade model) and using the individual energy trade flow (or the multiproduct trade model). Between 1995 and 2008, the average bilateral energy trade efficiency measured either by using the Malmquist index method for multi-product trade or by using the Malmquist index method for singleproduct trade has been increasing but their trends are different. In particular, the relative trade efficiency of the multi-product energy trade to that of the single-product energy trade declines while the standard deviation of estimated trade efficiency increases (Figure 6.4). This implies that bilateral trade efficiency, when flexible substitution between different energy products is allowed, is more likely to be diversified along with the increased mean. Table 6.3: Comparison of Energy Trade Efficiency, Single-Product Trade Multi-Product Trade Year No. of Obs. Mean Std. Mean Std Note : No. of Obs. = Number of Observations, Std. = standard deviation. Source: Authors own estimation. 151

19 Figure 6.4: Relative Trade Efficiency by Different Assumptions Mean and Standard Deviation Note : Relative STD = relative standard deviation. Source: Authors own estimation. In addition, the finding also shows that the exporters initial endowment in energy resources (among other controlled factors) also affects the possibility of bilateral trade creation in energy products. Efficiency of Energy Trade and Market Integration Based on the assumption of a multi-product trade and the imperfect substitution between different energy products, empirical results show that the average efficiency in bilateral energy trade across countries has been improving over time. Between 1995 and 2010, there are on average more than 14 percent growth in cross-country energy trade for every five years with constant income growth and natural (i.e., geographical or endowment) trade barriers, though the trend tends to decline over time. This finding reflects the globalisation and regionalisation throughout the world and their potential impact on EMI and in promoting bilateral/multilateral energy trade. 152

20 Table 6.4: Changes in Average Energy Trade Efficiency and its Components, Year Total trade Frontier movement Efficiency improvement Source: Authors own estimation. A decomposition analysis shows that the rapid increase in the bilateral trade potential of energy products is driven by two forces: the contribution of advanced countries efforts in further improving the trade efficiency, and the contribution of lagged countries efforts in catching up with advanced countries. On average, the advanced countries improving the trade efficiency accounted for around 70 percent of total efficiency gain in energy trade while lagged countries catching up with advanced countries accounted for around 30 percent of total efficiency gain. Figure 6.5: Trade Frontier Movement vs. Efficiency Gain, Source: Authors own estimation. 153

21 How does the trade efficiency of energy products change across different regions, in particular, within the EAS region? To answer this question, the bilateral trade flows were categorised into three groups: (i) the energy trade between EAS countries (intra-regional trade), (ii) the energy trade between EAS countries and the countries outside of the region, and (iii) the energy trade between countries outside of the region. The average efficiency of energy trade for each group of country pairs were estimated and presented in Figure 6.6. Figure 6.6: Comparison of Average Energy Trade Efficiency, by Country Groups, Source: Authors own estimation. Comparing across the three groups of countries, the average energy trade efficiency between EAS countries has been low relative to that of countries in other groups, but it increased quickly over time. The average energy trade efficiency between EAS countries has increased from 0.82 in 1995 to 0.89 in Over the same period, energy trade efficiencies between EAS countries and countries outside of the regions and that between countries outside of the region have declined from 0.88 and 0.87 down to 0.86 and 0.84, respectively. This implies that public policies aimed at improving EMI, among other factors, have played an active role in facilitating cross-country energy trade. Although the average energy trade efficiency between EAS countries has been increasing, there are still significant differences across countries. Figure 154

22 6.7 shows the average energy trade efficiency of three countries (the US, China, and Indonesia) in exports and imports. Over the period , energy trade efficiency of imports and exports between the US and its trading partners in the EAS region has been declining while that between China and its trading partners in the region has been increasing. This, in general, represents the changes in energy trade pattern between developed and developing countries due to their different performance in economic development and the related energy demand. As for Indonesia, energy trade efficiency of imports has been declining while that of exports has been increasing between 1995 and This finding is more likely to reflect the country s specific endowment in energy resources and its booming petrol and gas production. Figure 6.7: Average Energy Trade Efficiency of Imports and Exports: United States, China, and Indonesia A) Average energy trade efficiency in the United States 155

23 B) Average energy trade efficiency in China C) Average energy trade efficiency in Indonesia Source: Authors own estimation Implicit Share: Importance of Trade Components in Bilateral Energy Trade Using the Malmquist index to examine the gravity relationship between multi-product trade and its determinants, one can obtain additional results on the implicit prices for different trade components through the related 156

24 simulation. Usually, these implicit prices may reflect the relative importance of each energy products in the aggregate energy trade. Based on Coelli and Rao (2001), the simulation is used to derive the implicit prices of all three energy products specified in the model coal, petrol, and gas and the results are shown in Table 6.5. Table 6.5: Implicit Price of Coal, Petrol and Gas in Bilateral Trade Model Year ln_coal ln_petrol ln_gas Source: Authors own estimation. Between 1995 and 2008, implicit prices of petrol and gas have been increasing faster relative to the price of coal. The implicit prices of petrol and gas increased from 0.24 and 0.00 in 1995 to 0.39 and 0.01 in 2008 while that of coal declined from 0.41 in 1995 to 0.19 in This result partly reflects the increasing importance of trade in petrol and gas in total energy trade possibly due to changing preference. An important implication is to further improve the aggregate energy trade efficiency across countries, with more emphasis given to petrol and gas since their performance continues to increase over time. Policy Implication, Expected Result, and Future Development Study The development level of East Asia is vastly different from that of Cambodia, Lao PDR, Myanmar, and Viet Nam (also called CLMV countries). The 2008 gross national income (GNI) per capita in current value is US$630 for Cambodia, US$750 for Lao PDR, and US$910 for Viet Nam, while that in developed EAS countries, Australia has a GNI per capita of US$41,890, Japan has US$37,930, South Korea has US$21,570, and New Zealand has US$26,830, all in current values. The difference between the richest and the poorest countries is more than 60 times. Since narrowing development gaps is 157

25 a prerequisite for the process of regional integration, it is therefore very important to study the impact of EMI on growth convergence. It is widely believed that EMI will help participants to be more closely related through improving the bilateral trade efficiencies. Yet, how the trade creation process is achieved is not yet well understood. To address this issue, this study provides policy makers with some useful information on what kind of impact EMI can have on potential energy trade and the dynamic path of energy trade in different products, particularly on its impact on countryspecific products. As the analysis is narrowing the focus from the aggregate energy trade down to products, it improves the possibility of applying EMIoriented policies for the region and in trade-related countries. A few policy implications are expected. At the regional level, the productivity analysis will make it possible for stakeholders to understand the trade potential. This will help the regional policy makers to gauge their efforts. The estimated benefits will also reassure policy makers in their determination to move EMI forward. At the national level, first, information on the impact of EMI on product trade will help policy makers assess whether the consequence of EMI is acceptable since different kinds of energy products may have different strategic roles in each national economy. Second, this knowledge will make it possible for national policy makers to understand the impact by sector and, thus, they are able to formulate appropriate policies that will offset or enhance a particular impact. Conclusions This paper employs the Malmquist index approach to estimate the gravity relationship between bilateral energy trade and its determinants. Using a balance panel data of 40 countries covering the period between 1995 and 2010, a measure of energy trade efficiency at the aggregate level is provided and its change over time when considering the flexible substitution between different energy products, including coal, oil, and natural gas. Results show that along with the rapid growth in total energy trade, the trade efficiency in all energy products across countries have been increasing over the past two decades, particularly within the EAS region (though there are some cross- 158

26 country disparities). Both the advanced countries trade efficiency improvement and the lagged countries catch-up efforts played important roles in driving such a change. Results also show that different energy products contribute differently to the aggregate energy trade creation and to the corresponding trade efficiency gain. Generally, trade in coal accounts for the highest implicit prices but it has been declining over time relative to trade in petrol and gas, which suggests that trade in coal is losing its advantage over trade in petrol and gas. Thus, public policies that aim to improve regional EMI could benefit more by focusing on trade in petrol and gas. References Anderson, J. (1979), A Theoretical Foundation for the Gravity Equation, American Economic Review, 69(1), pp Anderson, J. and E. van Wincoop (2003), Gravity with Gravitas: A solution to the Border Puzzle, American Economic Review, 93(1), pp Coelli, T.J. and P. R. Rao (2001), Implicit Value Shares in Malmquist TFP Index Numbers, Centre for Efficiency and Productivity Analysis (CEPA) Working Papers No. 4/2001, Armidale: School of Economic Studies, University of New England. Costinot, A. and A. Rodriguez-Clare (2013), Trade Theory with Numbers: Quantifying the Consequences of Globalization, in A. Gopinah, E. Helpman, and R. Rogoff (eds.) Handbook of International Economics, Vol. 4, New York. Drysdale, P. and R. Garnaut (1982), 'Trade Intensities and the Analysis of Bilateral Trade Flows in a Many-Country World: A Survey', Hitotsubashi Journal of Economics, 22(2), pp Drysdale, P., Y. Huang, and K.P. Kalirajan (2000), 'Measuring and Explaining Trade Efficiency', a paper presented at the Asia Pacific Economies Program Seminar, Asia Pacific School of Economics and Management, The Australian National University, Canberra, August

27 Fare, R. S. Grosskopf, M. Norris, and Z. Zhang (1994), Productivity Growth, Technical Progress and Efficiency Change in Industrialized Countries, American Economic Review, 84, pp International Energy Agency (IEA) (2008), World Energy Outlook, Paris: IEA. International Energy Agency (IEA) (2012), World Energy Outlook, Jakarta: IEA. Kang, H. and M. Fratianni (2006), International Trade Efficiency, the Gravity Equation, and the Stochastic Frontier, Working Papers , Indiana: Kelley School of Business, Department of Business Economics and Public Policy, Indiana University. Kalirajan, K. (1999), 'Stochastic Varying Coefficients Gravity Model: An Application in Trade Analysis', Journal of Applied Statistics, 26(2), pp Kalirajan, K. and Findlay, C. (2005), Estimating Potential Trade Using Gravity Models: A Suggested Methodology, Tokyo: Foundation for Advanced Studies on International Development. Kalirajan, K. and K. Singh (2008), A Comparative Analysis of China's and India's Recent Export Performances, Asian Economic Papers, vol. 7(1), pp. 1-28, Cambridge: MIT Press. Kuosmanen, T., T. Post, and T. Sipilainen (2004), Shadow Price Approach to Total Factor Productivity Measurement: With an Application to Finnish Grass-Silage Production, Journal of Productivity Analysis, 22, pp Miankhel, A.K., S.M. Thangavelu, and K. Kalirajan (2009), Foreign Direct Investment, Export and Economic Growth in Selected Emerging Countries: Multivariate VAR Analysis, MPRA Paper No , Munich, Germany: University Library of Munich. Rose, A. (2004) Do We Really Know that the WTO Increases Trade?, American Economic Review, 94(1), pp Roperto, D. Jr. (2013), Trade Performance and Potential of the Philippines: An Application of Stochastic Frontier Gravity Model, MPRA Paper No , Munich, Germany: University Library of Munich. Roperto, D. Jr. and C. Edgardo (2014), Philippine Export Efficiency and Potential: An Application of Stochastic Frontier Gravity Model, MPRA Paper No , Munich, Germany: University Library of Munich. 160

28 Sheng, Y. and X. Shi (2013), 'Energy Market Integration and Equitable Growth Across Countries', Applied Energy 104, pp Shi, X and F. Kimura (2010), Energy Market Integration in the East Asia Summit Region: Review of Initiatives and Estimation of Benefits, Jakarta: Economic Research Institute for ASEAN and East Asia (ERIA). Subramanian, A. and S.J. Wei (2007), The WTO Promotes Trade, Strongly but Unevenly, Journal of International Economics, 72(2007), pp Wu, Y., X. Shi, and F. Kimura (2012), Energy Market Integration in East Asia: Theories, Electricity Sector and Subsidies, Jakarta: Economic Research Institute for ASEAN and East Asia (ERIA). Wu, Y., F. Kimura, and X. Shi, (2014), Energy Market Integration in East Asia: Deepen Understanding and Move Forward. Rutledge, Oxon, New York. Thanh, N. N. and K. Kalirajan (2006), Can Devaluation be Effective in Improving the Balance of Payments in Vietnam?, Journal of Policy Modeling, 28(4), pp